You can use Template Haskell to perform arbitrary computation at compile-time (even if it requires IO!), and then `lift` the result into a Haskell literal. This works for any type with a `Lift` instance (or with a bit of trick, any serializable type).<div><br></div><div>Coming back to your use case, you may try avoid using raw lists and switch to unboxed vectors, turn on -O2 and rely on stream fusion of the vector package. That will result in a considerable speedup.<br><br><div class="gmail_quote"><div dir="ltr">On Tue, Feb 27, 2018, 11:09 PM Dušan Kolář <<a href="mailto:kolar@fit.vut.cz">kolar@fit.vut.cz</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><u></u>
<div style="font-family:'Hack';font-size:9pt;font-weight:400;font-style:normal">
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">Dear Café,</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">   I'm trying to do very small, but impressive example about optimizations possible during Haskell compilation. So far, I can demonstrate that the following two programs (if compiled) perform computation in the same time:</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">1)</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">main =</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">    putStrLn $ show $ sum $ map (*(2::Int)) [(1::Int)..(100000000::Int)]</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">2)</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">main =</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">    putStrLn $! show $! sumup 1 0</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">sumup :: Int -> Int -> Int</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">sumup n total =</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">    if n<=(100000000::Int) then sumup (n+1) $! total+(2*n)</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">    else total</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">Nevertheless, I expect a question on comparison with C:</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">3)</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">#include <stdio.h></p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">int main(void) {</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">   long sum, i;</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">   sum = 0;</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">   for (i=1; i <= 100000000L; ++i) {</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">       sum += 2*i;</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">   }</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">   printf("%ld\n",sum);</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">   return 0;</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">}</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">Unfortunately, in this case the C is much more faster (it prints the result immediately), at least on my machine. Is it due to a fact that C compiler does a brutal optimization leading to compile-time evaluation, while ghc is not able to do that?</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">I'm using -O2 -dynamic --make ghc compiler flags. For gcc for C compilation just -O2, running Arch Linux.</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">Is there any option, how to force compile time evaluation? The reason, why I think it works this way is the fact, that when running out of long type values in C a code is generated that computes the values regularly (providing misleading value as a result) taking its time.</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">Best regards,</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px">Dušan</p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p>
<p style="margin-top:0px;margin-bottom:0px;margin-left:0px;margin-right:0px;text-indent:0px"> </p></div>_______________________________________________<br>
Haskell-Cafe mailing list<br>
To (un)subscribe, modify options or view archives go to:<br>
<a href="http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe" rel="noreferrer" target="_blank">http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe</a><br>
Only members subscribed via the mailman list are allowed to post.</blockquote></div></div>